Bayesian Causal Atlas · Vol. Obstetrics · Pearl Structural Causal Model

Pre-Eclampsia & DM199 / KLK1 Analogues — Structural Causal Analysis

Pre-eclampsia complicates roughly 2–8% of pregnancies and is a leading cause of maternal and perinatal mortality, with early-onset (<34 wk) disease carrying a long-term maternal cardiovascular-disease hazard of ~2.5–3.0. This oracle estimates the causal reduction in pre-eclampsia incidence (early-onset focus) across established prevention (aspirin, calcium, exercise) and investigational mechanism-targeted therapy — DM199 (recombinant human KLK1). Hazard ratios are from named trials; DM199 is shown as an explicit mechanistic projection (no outcome trial). For education, not individual medical advice.

Method. Structural Causal Model (SCM) with backdoor adjustment (Pearl). Interventions are not assumed independent: several act through the shared endothelial/placental-perfusion mediator, so their overlap is removed by an eigenvalue-corrected equicorrelation model at an adjustable mean cross-correlation ρ̄ (default 0.30). DM199’s direct KLK1→NO/PGI2 mechanism is separated as a gold direct path. Robustness to unmeasured confounding is quantified with the E-value. PN / PS / PNS under monotonicity. Every hazard ratio is cited; DM199’s figure is flagged as a projection, not a measured effect. The front door is resolved through an EXPLICIT mediator cascade (antiplatelet → endothelial / → metabolic → anticoagulation → disease state), not one lumped node: each intervention acts on a specific node, so same-node interventions are substitutes that saturate against each other, while different-node interventions are d-separated given the intermediate node and compose in series. The cross-correlation removal thus follows from the graph structure; the residual ρ̄ cleans up only the mediator-independent (direct) effects.
ρ̄ = 0.30
A–D (all)
A high (RCT/meta) · B cohort · C case-series/modelled · D consensus/provisional. Lower-grade interventions are excluded from the DAG, front-door pooling, Pareto, Monte‑Carlo & sensitivity.

Interventions

Tick the interventions to combine. Each shows its trial effect estimate, 95% confidence interval (CI), E-value, mechanism, and citation. ★ = in the current Pareto effective set but not yet ticked.

Combined causal estimate

Headline is the front-door estimate: shared Endo overlap removed via dose-response saturation; residual direct-effect overlap removed via the eigenvalue model at ρ̄.

1.00
Combined HR
0%
Relative risk ↓
Pooled E-value
Interventions selected (k)
0
Effective independent dimensions (n_eff)
0
Redundancy discount applied
0%
Baseline risk (illustrative anchor)
Absolute risk after intervention
Absolute risk difference (RD)
Backdoor-only HR (no front-door)
Endothelial/perfusion overlap removed
Number needed to treat (NNT)
ρ-sensitivity band (ρ 0 → 0.6)
Interpretation

Causal attribution

Under monotonicity + exogeneity (E-value bounds the exogeneity assumption).

Probability of Necessity (PN)
Probability of Sufficiency (PS)
Prob. of Necessity & Sufficiency (PNS, lower bound)
Causal DAG
Cross-correlation
Pareto (threshold)
Monte Carlo
Front-door mediation
What-if / If-not-for
Sensitivity
Antithesis

Faithful causal directed acyclic graph (DAG). Preventive interventions act largely through the shared mediator endothelial function / placental perfusion (spiral-artery remodelling, NO / PGI2 / EDHF), lowering placental-vascular resistance and thence pre-eclampsia incidence (Y), with a downstream edge to long-term maternal CVD. DM199 acts directly on the mechanism (recombinant KLK1 → bradykinin → NO / PGI2 / EDHF). Named confounders — pre-pregnancy metabolic/vascular risk, primiparity, prior PE, angiogenic balance (sFlt-1/PlGF) — open back-door paths (adjusted). Shared endothelial dysfunction confounds the PE→CVD edge. Mediator cascade: interventions attach to the node they act on (antiplatelet → endothelial / → metabolic → anticoagulation), which converge on the disease state and thence the endpoint — drawing the intermediate mediators explicitly is what exposes d-separation and per-channel saturation.

Confounders U:pre-pregnancy metabolic/vascular risk (obesity, chronic HTN, diabetes, CKD) · primiparity · multiple gestation · prior PE · genetic/angiogenic (sFlt-1/PlGF) → back-door paths (adjusted). Reverse pathway: shared endothelial dysfunction drives BOTH PE and later CVD (a collider/confounder for the PE→CVD edge).Low-dose aspirin(150 mg nocte)CalciumsupplementationStructured exercise(≥140 min/wk)L-arginine +antioxidantsMetformin(high-risk / obese)LMWH(prior placental PE)DM199 (rhKLK1)— candidateAntiplateletEndothelial /placental perfusionMetabolicAnticoagulationPre-eclampsiaMaternal /fetal outcomeFront-door: through endothelial function / placental perfusionDirect mechanism-targeted path (KLK1 → NO/PGI2/EDHF)Back-door confounding (adjusted)

Eigenvalue diagnostics for the selected interventions under an equicorrelation matrix (off-diagonal ρ̄). A large λmax relative to k signals redundancy; n_eff is the effective number of independent interventions actually contributing.

k (selected)
0
λmax
λmin
n_eff = (Σλ)² / Σλ²
Condition number

On mechanistic grounds ρ̄ ≈ 0.30 is defensible: aspirin, calcium, exercise, L-arginine and DM199 converge on the same endothelial/placental-perfusion mediator, so stacking them is partly redundant. ρ̄ is user-adjustable because their entry points differ (antiplatelet vs NO-substrate vs direct KLK1 vs smooth-muscle calcium), and LMWH sits largely off the effective pathway. Most of the overlap is now handled structurally by the mediator nodes (same-node substitutes saturate); ρ̄ governs only the residual correlation among direct effects.

Minimum-effective-set analysis. Set a target combined risk reduction; the model finds the smallest set of interventions — accounting for front-door mediator overlap — that reaches it, and highlights them. If the target exceeds what all interventions together can achieve, the full set is shown (never an empty one). "Apply" ticks exactly that set.

Target combined risk ↓ ≥ 50%

Monte Carlo propagation. Each selected intervention's log-effect is sampled from a normal distribution implied by its 95% CI; samples are combined with the same eigenvalue overlap discount. 5,000 draws.

Median combined HR
95% simulation interval
Standard deviation of combined HR
P(combined HR < 0.90)

Antithesis — challenging this oracle's own conclusions

DM199 has no outcome data — only surrogates. Every DM199 figure in this oracle is a mechanistic projection. The Phase-2 interim moved blood pressure and uterine-artery pulsatility (surrogates on the KLK1→NO/PGI2 pathway) with no placental transfer, but there is no pre-eclampsia-incidence, no fetal-outcome and no maternal-CVD trial. Its hazard ratio is an upper-bound hypothesis (grade D), not evidence, and the Phase-2 readout could fail to translate surrogate to clinical benefit.
Aspirin’s 0.38 is an early-onset, screen-enriched number. The ASPRE OR of 0.38 applies to preterm PE in women selected by a first-trimester algorithm; for all-PE and for term PE the effect is far smaller (~0.62 and near-null), and aspirin appears NOT to help pregnancies with chronic hypertension. Applying 0.38 to an unselected or late-onset population overstates benefit.
Calcium and L-arginine benefit are population-conditional. Calcium’s large effect is a low-dietary-intake phenomenon (RR ~0.55) that shrinks toward null when baseline intake is adequate (overall RR 0.80). L-arginine’s striking single-RCT result (RR 0.36) has not replicated cleanly; the pooled estimate (~0.66) is heterogeneous. These are not portable constants.
LMWH is mechanism-without-outcome. Heparin is biologically plausible at the maternal-fetal interface, yet the individual-patient-data meta-analysis found no significant reduction in recurrent placenta-mediated disease. It is retained as a near-null anti-target: a strong mechanistic story that the hard endpoint does not support.
The PE→CVD edge is confounded, not clean. Long-term maternal CVD after PE (HR ~2.5–3.0 for early-onset) comes from observational cohorts. Shared pre-pregnancy endothelial/metabolic dysfunction plausibly causes BOTH pre-eclampsia and later CVD, so treating PE does not automatically transfer to CVD reduction — the counterfactual rests on untestable no-unmeasured-confounding assumptions and is an upper bound.
Early-onset vs term PE are causally distinct. Early-onset (<34 wk) disease is placentation-failure dominated and is the DM199 / aspirin target; term PE is more maternal-constitutional. Effect sizes and mechanism weights differ by phenotype, so a single combined estimate blurs two different diseases.

What-if — the do-operator: P(Y | do(S))

Intervening on the selected set S with Pearl's do-operator (setting the interventions, not merely observing them). Contrast against do(∅) = no intervention.

P(outcome | do(∅)) — baseline
P(outcome | do(S)) — intervened
Absolute risk reduction (ARR)
Number needed to treat (NNT)

If-not-for — but-for counterfactual (leave-one-out)

For each intervention: "if not for this one, the combined front-door effect would be…". Isolates each intervention's marginal causal contribution after mediator-overlap removal, so shared-pathway agents are not double-credited.

If not for…HR without itHR with full setmarginal RRR lost

One-at-a-time sensitivity. Each intervention's effect is swung across its 95% confidence interval (others held at point estimate); the bar is the resulting swing in the combined front-door effect. A long bar means the combined estimate leans heavily on that single trial's precision.

Optimistic bound (all at CI-low)
Point estimate
Pessimistic bound (all at CI-high)
Pooled E-value (confounding robustness)

Front-door (mediation) decomposition. Each log-effect is split into an endothelial/perfusion-mediated (indirect) and a mediator-independent (direct) part. Indirect parts are pooled through the mediator with dose-response saturation, removing the mediator cross-correlation; direct parts keep the residual eigenvalue correlation at ρ̄. DM199’s effect is almost entirely mediator-directed by design — which is precisely why, with only surrogate data, a high mediated fraction paired with an unproven outcome is the graph’s central caution. Here mediated effects are pooled WITHIN each cascade node (dose-response saturation of substitutes) and composed in SERIES across nodes (d-separated channels), with the per-node reductions reported so the channel structure is visible.

Mediator saturation cap = 50% endothelial/perfusion gap closed
Sum of standalone Endo reduction (naive)
Combined Endo reduction after saturation
Mediator overlap removed (1 - saturation)
Direct-effect redundancy removed (1 - n_eff/k)
Front-door combined HR
Backdoor-only combined HR (comparison)
InterventionHR%Perf↑med-fracindirect logdirect log

Which % of cross-correlation is appropriate? Not one number. The mediator overlap is fixed empirically by the Endo saturation (currently removing of the summed mediated effect when interventions are stacked). Note a domain caveat: DM199 and LMWH illustrate the two failure modes — DM199 moves the mediator strongly on surrogates but has no outcome trial; LMWH has a mechanistic story but a near-null hard-outcome result. Neither should be read as an established preventive.

Front-door caveat (antithesis): textbook front-door identification requires an unconfounded mediator→outcome edge; here pre-pregnancy endothelial/metabolic dysfunction confounds both the mediator→PE and the PE→CVD edges, and effect sizes are strongly population-conditional (early-onset vs term; calcium-replete vs deplete). DM199’s figure is a projection from Phase-2 surrogates, not an outcome estimate.

Executive summary

Select interventions to generate a plain-language summary.